Analyzing the Spine Under Aircraft Ejection Loading
Dr. Ronald O. Stearman
Department of Aerospace Engineering and Engineering Mechanics
The University of Texas at Austin
May 6, 1998
This semester, Spinal Fusion attempted to conclude the spinal finite-element modeling project that began in the Fall of 1995. Specifically, the team attempted to recover and convolve all of the previous semesters' findings and data into a Finite-Element Model (FEM) in the I-DEAS software format. Though many methods of file retrieval and convolution were attempted, Spinal Fusion was unable to produce an FEM of the spine due to software and hardware difficulties. After the many failed attempts at model creation, Spinal Fusion was able to make recommendations towards the completion of this project and the production of an accurate and validated spinal FEM, thereby enabling the Air Force a quantitative evaluation of a pilot's flight status associated with the condition of his or her spine. Because of the inability to achieve the initial objectives of this project, the group created a new objective- an extensive literature review of the aircraft ejection and the spine. This review also included reviewing and consolidating the past groups' findings and references.
We would like to express our appreciation to several people. First, we would like to thank Dr. Stearman for his guidance. We would also like to thank Tim Ryan from the Anthropology Department for all assistance in retrieving the DigiBot images. Next, we'd like to thank Reuben Reyes and Scott Messac for all of the time and attention they afforded our group in the Aerospace Engineering and Engineering Mechanics Learning Resource Center (ASE/EM-LRC). Finally, we'd like to thank our families and loved ones for all of their support and guidance through the years (and the years to come).
Table of Contents
2.1.2Regional Characteristics of Vertebrae
2.4.2Injuries Associated with Ejection
2.5.1Swiss Foundation for Traffic Safety
2.5.2Brown University School of Medicine
2.5.7Articulated Total Body (ATB) Model
2.6Finite Element Modeling
3.1Fall 1995, Lewis and Grant
3.2Spring 1996, Bone Works
3.3Summer 1996, Hutchison/Littlefield
3.4Fall 1996, Spinal Tap
3.4.1FEM Development and Analysis
3.5Spring 1997, Bone Crusher
3.5.1I-DEAS FEM Development
3.5.2Bond Graph Model Development
3.5.3Strain Gage Testing
4.0Spinal Fusion Progress
4.3The I-DEAS Model
4.4.3Visualization Toolkit (vtk) Decimate
4.4.43-D Studio Max
5.13-D Studio Max Scheme
5.2.3Articulated Total Body (ATB) Model
Appendix A Request for Proposal from Brooks Air Force Base
Appendix B Material Properties Data referenced by Hutchinson/Littlefield
Appendix C Belytschko's Hypothesis
List of Figures and Tables
The dynamics of the human spine are difficult to quantify. Each spine is unique and reacts in a slightly different way than all other spines. Therefore, in order to study the dynamics of the spine, it is necessary to develop a model of the spine that is general enough to encompass the possible variations in form and function, but specific enough to react, under loading and stress, realistically.
The goal of this project is to develop a spine model that will be used to predict the loading and stress concentrations that a pilot can withstand during ejection. The loads and stresses will be applied to "healthy" and injured spinal models. The Request For Proposal (RFP) was submitted by Lt. Col. Drew, MD at Armstrong Laboratories at Brooks Air Force Base, who is responsible for certifying pilots for flight duty ; see Appendix A. Our project objectives were to create or modify an accurate I-DEAS model of the spine, to conduct finite element analyses on the "healthy" spine, and to repeat the analyses with a model modified to simulate injury.
Our three member team, Spinal Fusion, performed an extensive literature review in parallel with its attempts at producing a finite-element model of the spine. The team members did not perform tasks exclusive to their title; instead, each member contributed to all areas of the project. Jenna Bowling and Robin Kinsey concentrated on the research and documentation, while Tony Chao concentrated on file retrieval and manipulation. The entire group worked to bring the individual bones into I-DEAS.
This report is divided into four major sections: Background, Past Work, Spinal Fusion Progress, and Recommendations. The Background section gives a detailed explanation of the spine and its components, describes spinal injuries, details the fundamentals of aircraft ejections, and explains the current clinical findings on spinal injuries associated with ejection. The Past Work section summarizes the work completed by the five previous teams assigned to this project in terms of their computational and experimental findings, including any relevant recommendations. The Spinal Fusion Progress section discusses theory, the approach, and the results of the work pursued by Spinal Fusion. The Recommendations section outlines all of the suggestions for completion of the project.
This section details the basic concepts of this project. First, the structure and function of the spine are discussed, including modifications caused by spinal injuries. Next, aircraft ejections are described and diagrammed. Finally, the last section includes clinical findings on aircraft ejections and spinal characteristics associated with large accelerations and compression.
2.1 The Spine
The spine, or vertebral column, has several functions in the human body. It protects the central nerve, which runs through an opening in each of the interconnecting vertebrae. It also serves as the axial support for the skeleton and provides for the flexibility and bending of the back. The following two sections detail the general configuration of the spine and specifics of the spine according to region.
2.1.1 General Description
The spine consists of 33 vertebrae, in all, including those in the Sacrum and the Coccyx. The upper portion of the spine, called the Cervical region, is made up of
Figure 2.1 The Spine 
7 vertebrae (C1-C7); while the middle portion of the spine consists of 12 vertebrae (T1-T12) and is termed the Thoracic region. The next 5 vertebrae (L1-L5) make up the Lumbar region. The Sacral region and Coccygeal region are made up of 5 fused vertebrae (S1-S5) and 4 fused vertebrae (Co1-Co4), respectively. The configuration of the regions of the spine is illustrated in Fig. 2.1. The individual vertebrae of the five regions are labeled by the particular region and location within that region (e.g., the second vertebra of the Cervical region is C2).
Figure 2.2 Typical Vertebra 
Most of the load carried by the spine is supported by the vertebral body, which also serves as a resting-place for the intervertebral discs. Muscles and ligaments used in rotation and lateral flexion are attached to the spine by the transverse processes (2 per vertebra) which extend laterally from the point at which the pedicles and laminae are joined. The pedicles and laminae, shown in Fig. 2.2, connect the vertebral body and processes, forming the vertebral arch. This arch, together with the vertebral body, forms the vertebral foramen, which is the center hole in each of the vertebrae. The alignment of the vertebral foramina in the spine makes up the vertebral canal through which the central nerve runs .
Figure 2.3 Location of the Processes 
Including the two transverse processes, there are seven processes on a typical vertebra; see Fig. 2.3. These processes are either lever-like to enable the attachment of muscles and ligaments (e.g. the transverse processes) or articular (meaning joining) to help create bony joints . There are four articular processes on each vertebra, two facing up and two facing down. The articular processes of adjoining vertebrae form joints that limit movement of the vertebrae, thereby providing stability to the spine. The remaining vertebral process is the spinous process. The spinous processes vary in size, shape, and direction from one region of the spine to the next. Like the transverse processes, the spinous processes act as levers to which muscles attach. However, these muscles control posture and flexion/extension, lateral flexion, and rotation movements of the spine. The spinous processes are also the bones that can be felt as protrusions down the center of the back .
Figure 2.4 View of the Vertebral Interior/Exterior 
Each vertebra has a thin outer layer of compact (cortical) bone, which encases a soft, trabecular (cancellous) bone containing red bone marrow; see Fig. 2.4. The compact outer shell is thin on the discal surfaces but becomes thicker in the arch and its processes . In between each of the vertebra are intervertebral discs, which act as shock absorbers for the spine. Each disc is surrounded by a hard outer layer called the annulus fibrosus . Cartilage end plates positioned on the top and bottom of the disc act as a barrier between the disc and the vertebral body; see Fig. 2.5.
Figure 2.5 Vertebra/Disc Assembly 
The interior of the disc consists of a moist, jelly-like substance called the nucleus pulposus; see Fig. 2.6. The nucleus pulposus is cartilaginous and highly elastic with a high water content. It gets nutrients from diffusion of the blood vessels in the anulus fibrosis and the surface of the vertebral body. In general, the intervertebral discs allow free movement of the back by adding flexibility to the spine .
Figure 2.6 Disc and Spine Details 
Table 2.1 provides empirical quantities for the different material properties of each component of the spine. This table was extracted from a study of the lumbar region of the spine using a finite-element approach.
Table 2.1. Material Properties of the Spine 
2.1.2 Regional Characteristics of Vertebrae
The vertebrae in each region of the spine are characterized by differences in elements and geometry. When modeling individual vertebrae, it is important to acknowledge these differences. The following are some of the characteristics that distinguish one vertebra and one region from another.
Cervical Vertebrae: The vertebrae in the cervical region of the spine are characterized by a foramen in each transverse process of C1-C6. These foramina transmit vertebral blood vessels, which supply the superior spinal cord and the posterior brain. Furthermore, each of the articular processes is angled to permit flexion, extension and rotation of the head.
Figure 2.7 Cervical Vertebra 
The C3-C6 vertebrae are called typical vertebrae. They have a triangular shaped vertebral foramina, vertebral bodies which are wider from side to side than from front to back, and forked and short spinous processes. The C1, C2, and C7 vertebrae are called atypical vertebrae. The atlas vertebra, C1, lacks a vertebral body. Instead, it has an anterior arch, which has a facet joining C1 to the dens of C2. C1 also lacks a spinous process. The axis vertebra, C2, is characterized by its odontoid process, or dens, which is a tooth-like structure joining the body of the axis to the anterior arch of the atlas. The C7 vertebra is distinguished by its very long spinous process . The general characteristics of the cervical vertebra can be seen in Fig. 2.7.
Figure 2.8 Thoracic Vertebra 
Thoracic Vertebrae: The T1-T10 vertebrae of the thoracic region have a facet on the transverse processes, which joins with the tubercle rib. Their vertebral foramina are circular-shaped, and their vertebral bodies are heart-shaped. The thoracic vertebrae also have very long, slender spinous processes, which are directed downward . The general characteristics of the thoracic vertebra can be seen in Fig. 2.8.
Figure 2.9 Lumbar Vertebra 
Lumbar Vertebrae: The vertebrae in the lumbar region of the spine are characterized by their massive, kidney-shaped vertebral bodies and missing costal facets. Their vertebral foramina vary from oval-shaped to triangular-shaped, and their spinous processes are short and oblong . The general characteristics of the lumbar vertebra can be seen in Fig. 2.9.
Figure 2.10 Sacral Vertebra 
Sacral Vertebrae: The sacral vertebrae are fused together to form the sacrum. The anterior surface, representing fused vertebral bodies, is smooth and concave; the posterior surface, representing fused vertebral arches, is rough and convex. The posterior surface is also distinguished by its media sacral crest, which is just the fusion of the spinous processes of S1-S4. The S5 vertebra lacks a laminae and a spinous process . The general characteristics of the sacral vertebra can be seen in Fig. 2.10.
Figure 2.11 Coccygeal Vertebra 
Coccygeal Vertebrae: The Co1-Co4 coccygeal vertebrae consist of vertebral bodies only and are fused together. They provide attachment for pelvic muscles and ligaments and do not bear weight . The general characteristics of the coccygeal vertebra can be seen in Fig. 2.11.
2.2 Spinal Injuries
The spine can become injured in a variety of ways. Spinal injuries can be the result of a sports-related occurrence, the lifting of a heavy object, an auto accident, or ejection from an aircraft. Ultimately, the spine sustains an injury due to the stresses imposed upon it becoming greater than the material strength of the spinal components. Several examples of common problems or injuries are described by Theresa McIntosh :
Figure 2.12. Disc Herniation 
2.3 Spinal Treatments
When a spinal injury occurs, there are many types of non-surgical treatments that can be applied, such as medication, physical therapy, or injection therapy. Surgical options are then considered as a last resort for treating a spinal injury. One surgical solution for a herniated disc is a disectomy. A disectomy is an operation performed to remove the herniated portion of the disc, thereby relieving any pain associated with the ruptured disc. People with spinal stenosis can receive a laminectomy, which is an operation used to remove the vertebral lamina in order to enlarge the opening through which the spinal cord and spinal nerves run .
Resulting from a disectomy, there may be abnormal movement between the two adjacent vertebrae. This is due to the missing intervertebral disc that had acted as a cushion between the bony vertebra. In this situation, a procedure called spinal fusion can be implemented to fuse the two vertebrae together. This is accomplished by surgically applying bone graft and/or spinal instrumentation. The spinal instrumentation would be used to stabilize the graft or fixate the bones while a solid bone mass forms. The bone graft may be an autograft, bone taken from your own body, or an allograft, bone obtained at a bone bank. Instrumentation applied in spinal fusion in the cervical region may consist of plates along with screws, wires, or bone plugs for bone graft. Thoracic and lumbar spinal fusion may utilize rods, hooks, screws, wires, and bone grafts .
2.4 Aircraft Ejection
Aircraft ejection is a necessary life-saving event for a pilot. Ejection from an aircraft occurs at the pilotís command and is usually a last resort for the pilot. Aircraft ejection seats have been used over 12,000 times to date . The following sections detail the typical ejection sequence, characterize the forces on a pilot through ejection, and describe the common injuries associated with ejection. Figure 2.13 is an example of a ground-based test ejection .
Figure 2.13. Ground Ejection Test 
The typical ejection sequence can be shown using a description provided by The Ejection Site, a web page devoted to ejection information. This page details the ejection sequence of the ACES II ejection seat. The ACES II model is an ejection seat produced by the McDonnell-Douglas Aircraft Corporation and is a typical representation of aircraft ejection seats used by the United States Air Force .
Figure 2.14. ACES II Event/Time Sequence 
Figure 2.14 shows the sequencing of the ACES II ejection. As can be seen in the figure, there are three modes of operation, and each mode is employed dependant upon velocity and altitude of the aircraft . Mode 1 is a low velocity (less than 250 knots)/low altitude ejection; Mode 2 is a medium speed/high altitude (above 15000 feet) ejection. Mode 3 is a high speed/high altitude ejection, which is the most dangerous ejection for two reasons: low levels of oxygen in the air at high altitudes and the speed at which the pilot is traveling upon exiting the aircraft.
Figure 2.15. ACES II Mode 2 Ejection Sequence 
Figure 2.15 shows the typical Mode 2 ejection sequence, where the drogue system deploys. The drogue system consists of one "hemisflow chute, a small extraction chute, and the drogue mortar. The drogue mortar is fired in Mode 2 and Mode 3 to slow and stabilize the seat-man package. This is intended to prevent or limit the injuries to the crewmember as he/she is exposed to the windblast after exiting the aircraft" .
One important aspect of ejection is the time-period of ejection. In as little as 5.5 seconds (for this particular model of ejection seat), the pilotís spine has gone through a series of compressions and "jerks" that can cause major injuries . These injuries can occur as fractures of the bones in the spine or as damage to the intervertebral discs.
2.4.2 Injuries Associated with Ejection
The ejection from an aircraft produces high G-forces on the pilot and results in compression loading to the spine. Many pilots sustain injuries in the cervical region of the spine. These injuries are due to the position of the head and the resulting "jerk" in the neck from the impact of ejection. Alternatively, pilots who have had previous injuries in their lower spine are more susceptible to fractures in the thoracic and lumbar regions of the spine.
In a report sponsored by the Chief Bureau of Medicine and Surgery, Navy Department, Clinical Investigation Program, demographic studies of Navy pilot ejection during Operation Desert Storm were performed in order to determine the relationship between vertebral fracture, flight parameter and ejection; see Table 2.2 .
Table 2.2. Demographics from Desert Storm Pilots 
The table shows that all of the vertebral injuries occurred in the thoracic and lumbar regions of the spine, and the highest incidence of injury occurred in high speed, low-altitude attack missions. This ejection study produced demographic data, which can be used in FEM verification.
The injuries associated with helicopter crashes were studied by Simula, Inc. under contract from the Safety and Survivability Technical Area of the Aviation Applied Technology Directorate, U.S. Army Aviation Systems Command. In this study, cadavers were subjected to load limit testing . These tests produced experimentally determined curves for ultimate compressive strength of the vertebrae. These results are shown in Fig. 2.16, and will be helpful in FEM validation.
Figure 2.16. Ultimate Compressive Loads in the Vertebrae 
2.5 Additional Research
Many other universities and institutions around the world are producing finite element models of the spine for running various analyses. For example, there are FEMs being created to study the effects of automobile accidents on the spine (e.g. whiplash), the benefits of using a helmet in football to help reduce head and spinal injuries, etc. The remainder of this section provides examples of FEMs constructed for a variety of motives. The current section also takes a general look at several simulator models being used for various studies of/with the human body.
2.5.1 Swiss Foundation for Traffic Safety
The Swiss Foundation for Traffic Safety is supporting a study by the Biomedical Engineering and Medical Informatics Laboratory in Zurich, Switzerland, in which they are working to develop an FEM of the cervical spine to be used in an analysis of the cervical spine during a rear end impact . They have taken MRI-scans of the cervical spine of a volunteer and used them to model the correct geometry of the cervical spine. The Instituteís initial plan is to join its model with an existing rigid-body model of the rest of the human body, which can then be used to simulate test conditions normally used in tests requiring real volunteers.
2.5.2 Brown University School of Medicine
An example of a sports-related injury study is the study by the Department of Orthopedic Surgery and Division of Sports Medicine at Brown University School of Medicine in Providence, Rhode Island . This study will determine the effect of protective football equipment on the alignment of the intact lower cervical spine and on the alignment of a partially destabilized C5-6 motion segment. Fifteen cadavers were used in the study, and for each specimen, four lateral cervical radiographs were obtained.
The Test Device for Human Occupant Restraint (THOR) is an advanced frontal crash test dummy supported by the Research and Development Office at the NHTSA Vehicle Research and Test Center . THOR was created as an "effective tool for whole-body trauma assessment in a variety of automotive occupant restraint environments." The THOR test dummy has many advanced features that has enabled it to be used in testing for vehicle occupant safety by auto manufacturers such as Ford, GM, and Chrysler. Two THOR prototypes have been constructed and are currently used in various international tests. The THOR system has a working mechanical "spine" included in the hardware that attempts to simulate realistic human movement.
2.5.4 Transom Jack
Transom Jack is produced by Transom Technologies, Inc., a privately owned company headquartered in Ann Arbor, Michigan . Transom Jack is a "real-time visual simulation that enables you to create virtual environments by importing CAD data or creating objects, populate an environment with biomechanically accurate human
Figure 2.17 Transom Jack 
figures, assign tasks to these virtual humans, and obtain valuable information about their behavior." The Transom Jack software enables the user to input physical characteristics such as weight, height, etc. The Transom Jack human figure is a complex, realistic simulation of a human. It is made up of 74 segments, 73 joints, a 22-segment spine, and 150 degrees of freedom. Transom Jack can be used in a variety of simulations which may be desired by a company, for example, to train employees or to verify a design. Some examples of companies who have utilized Transom Jack are NASA, Lockheed Martin, and the U.S. Air Force, Army, and Navy. An example of the Transom Jack model is shown in Fig. 2.17.
The Advanced Dynamic Anthropomorphic Manikin (ADAM) was initiated by Armstrong Laboratory in response to the Crew Escape System Technology (CREST) Advanced Development Program Office . CREST desired an advanced manikin for ejection system testing. The objectives of Armstrong Laboratory were to validate ADAM for use in testing escape and protection systems, to modify manikins in order to represent female pilots, to use measured manikin response data to determine the chance of human injury, and to create databases of manikin and aircraft system interface properties which can be referenced to ensure safe designs and to validate computer simulations.
The ADAMS/Android is a software component designed by Mechanical Dynamics, Inc. which is used in their Automatic Dynamic Analysis of Mechanical Systems (ADAMS) suite to model the human body. "The software enables users to easily and dynamically create realistic human models and then use ADAMS to study dynamic interaction of complex human/machine systems."
Figure 2.18 ADAMS/Android 
The android model can be manipulated within an environment in a number of ways, and the posture and individual body segments can be altered to produce a variety of test simulations. One additional feature of this modeling package is its ability to interface directly with the I-DEAS software package. Samples of potential applications of this modeling package are accident reconstruction, occupant protection studies (including aircraft ejection), and zero-gravity environment studies. An example of the ADAMS/Android is shown in Fig. 2.18.
2.5.7 Articulated Total Body (ATB) Model
The Articulated Total Body (ATB) model was developed by Armstrong Laboratory at Brooks Airforce Base to predict human body dynamics in hazardous events, such as crashes and aircraft ejections . It is a microcomputer based software package based on the Crash Victim Simulator (CVS) developed by the National Highway Traffic Safety Administration (NHTSA) during the early 1970's.
Figure 2.19 Articulated Total Body Model 
A more user-friendly version of the ATB called DYNAMAN was completed in 1992 and can run under DOS and on Silicon Graphics Workstations. An ATB user's group meets annually to share modeling experience and additional products. The latest version of the ATB is version V.1, which was introduced by the Air Force Research Laboratory at the 4th Annual ATB User's Conference on April 30, 1998. An example of the ATB model is shown in Fig. 2.19.
2.6 Finite Element Modeling
Spinal Fusionís ultimate goal was to develop an accurate Finite Element Model (FEM) of the spine for simulating aircraft ejection loads to a spine that has undergone spinal fusion. Though this objective was not realized, Spinal Fusion was able to identify a plan for actual completion of the project. The following section provides the theory behind finite-element modeling and background information on specific FEM software tools, as summarized by the Bone Crusher group from the Spring of 1997.
Finite element modeling is the modeling of a continuous system, which has an infinite number of degrees of freedom, using a representative geometry of that system made up of a finite number of smaller elements and node points. The more elements in the model, the more accurate it is. The material properties, displacements, and other system characteristics are represented by mathematical functions between nodes. This finite element model can then be used to determine the stress, strain, and displacement of the structure resulting from external loading .
Simple shapes, such as triangles or squares, are used to form the geometry of the model. The points at which these shapes, or elements, are connected are the node points. The elements can be connected in any manner necessary to yield the optimum model. Once the finite element model has been created and the system characteristics have been established in the model, a global stiffness matrix can then be formed for the whole structure. Given the forces and boundary conditions, the unknown displacements at each of the node points can then be used to determine the stresses and strains acting on each element .
2.6.2 Software Packages
The finite element modeling and analysis would be extremely time-consuming without appropriate computer software. Several software packages have been used in the attempt to develop a finite element model for this project. The following summaries were obtained directly from the Spring 1997 Bone Crusher Final Report .
I-DEAS: The Integrated Design Engineering Analysis Software (I-DEAS), is a package of mechanical engineering and structural analysis applications. The I-DEAS program consists of various "Applications", which are further subdivided into various "Tasks". The applications are design, drafting, simulation, test, manufacturing, management, and geometry translators.
The I-DEAS package consists of multiple windows and a user-friendly graphical interface with a variety of well-defined task icons. I-DEAS includes a unique filtering tool that allows the user to turn off certain parts of a model and only display what is currently needed.
ABAQUS: ABAQUS is a multi-purpose finite element analysis program that has the capability of performing most dynamic and static structural analyses. Unfortunately, the ability to create complex models is limited by the fact that ABAQUS is driven by script files. These script files must be manually created, so in order to make an accurate model of the spine the geometry of each vertebra would have to be manually entered. Overall, the program comes in tree separate modules: ABAQUS/Standard, the finite element creation tool, ABAQUS/Explicit, which is used for explicit dynamic analysis, and ABAQUS/Post, which provides a graphical representation of the finite element solutions which are run.
Pro/Engineer: Pro/Engineer is primarily a computer aided design and modeling tool. Included within Pro/Engineer is a simplified finite element modeling capability, however the ability to perform actual structural analysis is quite limited.
DIGIBOT: The DIGIBOT 3-D Laser Digitizer is maintained by the UT Anthropology Department and is used to scan in the vertebral geometry. The DIGIBOT uses laser-ranging technology to create accurate representations of the vertebral geometry. The scans are created by mounting the vertebra on a scanner platter and selecting the desired scanning parameters. The scanner allows for the selection of the translational and rotational spacing to define the resolution of the scan. Currently, all or most of the vertebra have been scanned in and are maintained by Tim Ryan, a graduate student in Anthropology.
In addition to the above summaries, Spinal Fusion investigated other software packages, as described below.
DigiEdit: DigiEdit is an editing tool that enables modification of the scanned in DIGIBOT data files. DigiEdit provides a 2-D window for viewing and modifying individual cross sections and a 3-D window for viewing the set of cross sections. It also has a Triangulator that joins points between adjacent cross sections. The images of each of the vertebra have already been edited with DigiEdit by previous groups , though it was necessary to re-edit several of the bone images to achieve a complete surface mesh. The surface meshes were then saved in object (obj) file format in order to be edited in SCULPT.
SCULPT: The DigiEdit images were highly detailed and too complicated for the needs of this project. Therefore, it was necessary to decimate the DigiEdit image files using a modeling tool to yield a more simplified geometric model in I-DEAS. SCULPT was successfully utilized in the past for this decimation purpose  and was used by Spinal Fusion to reduce the complexity of the bone surface meshes by about 70 percent. After decimation, the files were exported from SCULPT in the dxf file format.
FORMZ: FORMZ is a modeling tool that enabled the group Spinal Tap to save the vertebrae images in the IGES format. This format could then be imported into I-DEAS using the I-DEAS Geometry Translator module. For previous groups, this was the only successful way to import the vertebral images into I-DEAS. This software package was no longer available for use by Spinal Fusion.
AutoCAD 14: AutoCAD 14, a Computer Aided Drafting software package, was used to view and export the dxf format bones into the IGES (igs) file format using the igesout command-line function. Dxf is the abbreviation of Data Exchange File, a two-dimensional graphics file format supported by virtually all PC-based CAD products. It was created by AutoDesk for the AutoCAD system. IGES is the acronym for Initial Graphics Exchange Specification, an ANSI graphics file format for three-dimensional wire frame models.
3-D Studio Max: 3-D Studio Max is a software package generally used to create animations and movies with a professional look in multiple perspectives. This software package was used by Spinal Fusion to first import the bones in dxf format, to then manipulate and position the bones in the spinal curvature, and to finally export the entire spine in dxf file format. This process was vital, because dxf was the only file format that Spinal Fusion could translate directly into the igs format (using AutoCAD 14) that I-DEAS could then import.
Visualization Toolkit Decimator: Vtk Decimate by General Electric Computer Graphics and Systems Programs was explored as a possible alternative to SCULPT for decimation. This program, however, corrupted the bone surface meshes, so was not pursued.
StlUtil: This utility program designed by B. Michel for stereolithography files was used to translate files between several different file formats, though the final scheme used by Spinal Fusion bypassed this program.
It should be noted that the above single-spaced summaries are quoted from the Bone Crusher Final Report. Spinal Fusion did not actually need to use all of these software applications, though each package should be considered for future work in this area.
3.0 Past Work
The investigation by the University of Texas Aerospace Engineering Department of loading and stressing of bones began in the Fall of 1995. Five groups have each spent a semester on this study. The following is a summary of the past work by these groups. The last group, Bone Crusher, performed an extensive literature review of all previous work completed. A short summary of their review can give one a better appreciation of the "whole picture" of the project. The following sections outline the basic findings of each group.
3.1 Fall 1995, Sharon Grant and Kari Lewis 
In fall 1995, the Grant/Lewis team attempted to develop a Finite Element Model (FEM) of the tibia and experimentally verify the model through photoelastic testing.
3.1.1 FEM Development
Grant and Lewis began the FEM development by using a DIGIBOT scanner to generate a surface image of the tibia. This image was imported into DigiEdit to generate the inner-surface of the compact bone shell and to facilitate the editing of both ends of the tibia. The cross-sectional data from DigiEdit was then imported into Excel, processed, and imported into ABAQUS. An ABAQUS script was generated to organize the coordinate geometry, define the material properties, constrain the geometry, and apply the loads. The resulting output file was analyzed by ABAQUS-Post. Grant and Lewis used a transverse lyisotropic material definition of the compact bone, but they ignored the trabecular bone. The articulating ends of the tibia were also ignored.
3.1.2 Photoelastic Testing
Their photoelastic test results were inconclusive for several reasons. The specimen was poorly aligned for vertical loading, and the photoelastic coating had poor adherence to the specimen. The tibia also had excessive adhesive between the photoelastic material and the bone.
Grant and Lewisí FEM of the human tibia was plagued by high level of element distortion. The photoelastic tests did not agree with their computational results. These results actually have no real bearing on Spinal Fusion's modeling process, but do give background on the University of Texas involvement with this project.
3.2 Spring 1996, Bone Works 
The members of Bone Works group were Judson Frieling, Adam Richards, and Jonathan Mead. Their goals during the semester were to improve on Grant and Lewisí FEM, or if this failed, to develop a new FEM and verify it through photoelastic testing. At first, Bone Works tried to correct the Grant/Lewis model by manually patching the FEM. This procedure involved manually rearranging the cross-sectional layers of the FEM with high distortion angles and removing those particular cross-sections. This attempt failed due to excessive distortion.
3.2.1 FEM Development
Bone Works then tried to build a more accurate FEM. The group imported data generated from DIGIBOT and DigiEdit into AutoCAD. AutoCAD was used to generate cross-sectional planes of data, which was imported into Excel. Closed- spline curves defining the cross-sectional contours were generated in Excel. The coordinates were further processed by Excel, and then imported into ABAQUS. Finally, ABAQUS and ABAQUS-Post were employed to analyze the data.
3.2.2 Photoelastic Testing
Bone Works believed that the photoelastic adhesive that Grant and Lewis used was too poor to yield accurate results. Moreover, Bone Works concluded that white light should be used instead of monochromatic light because white light gives up to 12 bands of color for each order. Bone Works found it difficult to produce results from compression tests (loads up to 100 pounds), and further research was suggested.
The FEM created by Bone Works was distortion-free, resulting in considerably fewer stress concentrations than the Grant/Lewis model. Uncertainties remained in the Bone Works model due to inaccurate material property definition. Lastly, the photoelastic experiments failed to verify the computational model. Again, the results obtained in this study have no direct bearing on the ejection loading problem, but do give useful background information.
3.3 Summer 1996, Hutchison/Littlefield 
In summer 1996, the focus of FEM analysis shifted from the tibia to the vertebral column in response to a Request for Proposal from Brooks Air Force Base. Brooks AFB was interested in a FEM of the spine that could determine the stresses induced in a previously injured spine during pilot ejection. This projectís objective was to find out whether a pilot with a spinal injury was fit for ejection. The team members for this semester were William Hutchison and Brian Littlefield.
3.3.1 FEM Development
Hutchison and Littlefield began by importing data from the DIGIBOT images of the C1-C3 vertebrae into ALGOR V. After an internal mesh was generated by ALGOR V, a PERL script was written to translate the data for ABAQUS to read. The model was then analyzed by ABAQUS-Post. However, the model, having about 30,000 nodes, was too complicated to be processed. Hutchison and Littlefield then simplified the vertebral body by modeling it as a cylinder; see Fig 3.1. A. Shirazi-Adlís study was
Figure 3.1 Hutchinson/Littlefield Simple Model 
referenced for the material properties of the lumbar region; see Appendix B. The groupís model was also derived from Belytschkoís study; see Appendix C.
Highly accurate (and complicated) surface meshes of the C1 through C3 vertebrae were created by Hutchison and Littlefield. Because of limited computing resources, a simplified FEM of the spinal column was built for future analysis.
Spinal Tap consisted of Carla Haroz, Jeremy Jagodzinski, Robert Rose, and Vladmir Sierra. Their goals included developing an accurate and simple FEM for analyzing and investigating the validity of photoelastic and thermoelastic tests on the tibia.
3.4.1 FEM Development and Analysis
Like the previous group, Spinal Tap employed DIGIBOT to generate a surface mesh of a C1 vertebra. Each vertebra was triangulated by DigiEdit to form a solid model. Then Sculpt was used to decimate by half the number of vertices and facets by removing patches over vertices that did not contribute to the curvature of the surface. Thus, the model was simplified without sacrificing accuracy. Next, FORMZ was used to link the output from Sculpt to I-DEAS. In I-DEAS, an intervertebral disc was modeled as an ellipsoid, which was molded by the surfaces of two vertebral bodies by using Boolean operations; see Fig. 3.2. After the vertebral bodies and the intervertebral disc were in place, four loading configurations were simulated.
Figure 3.2 Spinal Tap's Vertebra/Disc Assembly 
3.4.2 Photoelastic Testing
Like its predecessors, Spinal Tap attempted to use the tibia as a specimen for photoelastic testing. The problems Spinal Tap encountered were similar to those of the previous groups, related to the thickness and uniformity of the adhesive and the photoelastic film and delamination of the adhesive when loads were applied. Spinal Tap concluded that the photoelastic measurement systems were not appropriate for this project.
3.4.3 Thermoelastic Testing
SPATE (Stress Pattern Analysis by Thermal Emission) was chosen for thermoelastic testing. This technique employed an infrared scanner to detect heat that emanated from the surface of the specimen. This heat measurement was used to determine the amount of surface strain. A load range of 120 to 290 lbs at a frequency of 3 Hz was used.
The most successful result was the creation and simulation of the FEM model of the intervertebral disc. Four loading conditions were applied, and the results were in agreement with several of Belytschkoís hypotheses; see Appendix C. However, both photoelastic testing and thermoelastic testing did not yield meaningful data. Photoelastic testing failed because of physical problems similar to those that previous experimental groups encountered, and thermoplastic testing failed mainly due to insufficient heat generation from the tibia surface.
3.5 Fall 1997, Bone Crusher 
The members of Bone Crusher were Cathy Cantu, Luis Frias, Eric Lehman, and Jeremy Rust. Bone Crusher's initial objectives were to conduct FEM analysis of the spine using an I-DEAS model and perform experiments on a tibia to verify the FEM analysis results. Another purpose of mechanical experiments on the tibia was to determine if the vertebra could be modeled as a homogeneous material with the estimated material properties. However, due to the expiration of the I-DEAS package in the Learning Resource Center during the middle of the semester, Bone Crusher was unable to complete their I-DEAS model. The group turned to Bond Graph modeling as an alternative means of analysis.
3.5.1 I-DEAS FEM Development
Bone Crusher initially attempted to follow the recommendations of Spinal Tap in the development of the FEM model. The group was successful in scanning the complete spine using DIGIBOT, but failed to import the data into a format readable by I-DEAS. Since scanned-in images of the spine were not available for I-DEAS, Bone Crusher made the bold assumption that the vertebrae can be represented by a much simpler mass/spring/dashpot system. This approach was first proposed by Williams and Belytschko in 1982; see Appendix B. The model that Bone Crusher constructed previous to I-DEAS' expiration consisted of a simple column arrangement of beam elements representing the 12 vertebrae of the thoracic region and the 5 vertebrae of the lumbar region. Bone Crusher confessed that the model was an oversimplification of the spine, since the muscles and ligaments were not included in the model. Spinal Fusion was unable to locate this model in the Bone Crusher computer file archive.
3.5.2 Bond Graph Model Development
The Bond Graph Model of the spine was essentially a system of masses, springs, and dashpots. The masses represented the vertebral bodies, bone muscles, and bone organs, whereas the intervertebral discs were modeled by springs and dashpots. With the help of Tom Connolly, a mechanical engineering graduate student at UT Austin, Bone Crusher was able to translate the system model into a bond graph model. A MATLAB script written by Connolly was then used to generate intervertebral disc compression graphs.
3.5.3 Strain Gage Testing
Strain Gages were placed on a tibia specimen to measure the strains at two perpendicular orientations at different locations on the tibia. Axial loads from 0 up to 200 lbs were applied. The data were translated into Von Mises Stress along the surface of the tibia to validate the hypothesis that the material could be estimated as homogeneous.
Since the I-DEAS model of the spine was incomplete, quantitative results were not presented. In addition, because the I-DEAS model cannot be located, Spinal Fusion was unable to follow the Bone Crusher recommendations for using the simplified model. The Bond Graph Model, on the other hand, produced compression versus time graphs for each of the 23 intervertebral discs. Data compiled from the strain gage experiments on the tibia was insufficient to establish the hypothesis of homogeneity or validate the computational models made by Bone Crusher.
4.0 Spinal Fusion Progress
During the semester, Spinal Fusion created a large literature review, as well as tried to recover the old models in order to modify and validate them. The following sections describe Spinal Fusionís attempts at data recovery and our progress towards the completing the recommendations made by the former Spinal FEM group Bone Crusher. Included below is a description of the Spinal Fusion Design Team Home Page that was created to inform others of our progress and allow for easy group communication.
4.1 Data Recovery
The computer files used by Spinal Tap (Fall í96) and Bone Crusher (Spring í97) were located and explored. Table 4.1 shows the type of file, location (site), and login/password information.
Table 4.1. File Locations 
Triangulated IGS Files
Spinal Tap Intervertebral Disc
Bone Crusher Files
The files found in Bone Crusherís directory contain the I-DEAS Version 4 simplified model of spine that was created by the Hutchinson/Littlefield group. After being duplicated on a zip disk for backup purposes, these files were converted to I-DEAS Master Series 5.
The files recovered from Spinal Tapís group contain the I-DEAS version 4 model of an intervertebral disc created from the surface topology of a C1 vertebrae. After being duplicated on a zip disk, these files were also converted to I-DEAS version 5.
The account in which the triangulated IGS files were located is a shared File-Transfer-Protocol site for the University of Texas at Austin Anthropology Department. These files were not usable in I-DEAS Master Series 5, for unknown reasons. It was necessary to re-archive these files by returning to the original DigiBot scans, re-editing them, and exporting those files through DigiEdit.
4.2 Web Page
A web page that includes the most important research links and updated facts was created by Spinal Fusion at the beginning of the semester and was updated throughout the semester. These links include the midterm report references and the ASE 363Q web page. Also included on the web page are the email addresses of each team member, including the page creator Jenna Bowling. This web page can be accessed directly at http://www.ae.utexas.edu/courses/spine.
4.3 The I-DEAS Model
Due to the premature departure of a Bone Crusher team member last year, they were unable to complete an accurate quantitative finite element model of the spine. Towards the completion of an FEM model of the spine that will accurately simulate the stress and displacement distribution in the spine, they made the following recommendations:
Enhance and implement Spinal Tapís FEM of the intervertebral disc, convert the IGES scanned-in images or DIGIBOT files to a format that can be read by I-DEAS, include boundary conditions, input material properties, and replace the spring-dashpot model of the intervertebral disc with elements.
The enhancement and implementation of Spinal Tapís FEM of the intervertebral disc is important in modeling an injured intervertebral disc. The I-DEAS FEM element thickness needs to be updated to create a more representative slice of the disc. Also, the material properties for the various portions of the disc have been found and the method of creating different properties for different elements established. The properties must simply be entered and various cases run. In addition, the FEM of the intervertebral disc may be used to model injury. Injury can be simulated by degenerating the intervertebral disc. This involves increasing the stiffness of the springs (annulus fibrous) and decreasing the damping constant (nucleus pulposus). Replacing the spring-dashpot intervertebral disc model with elements will more accurately model its behavior. Such a model would reveal the stresses in the intervertebral disc. That actual creation is rather simple, once the laminate properties are entered.
It should be noted that these recommendations are quoted from the Bone Crusher Final Report .
Spinal Fusion could not follow these recommendations because we were unable to find the Bone Crusher model. Without a model to alter, Spinal Fusion decided to create a new model using the combined recommendations of all of the previous groups.
4.4 New Model
Several schemes for model creation were developed using the recommendations of the past semesters. Eventually, each of these schemes was found flawed for different reasons. Each scheme began by using the original scans from DigiBot and editing them in DigiEdit. The following sections outline each scheme developed and attempted by Spinal Fusion to create a new model.
4.4.1 Direct Method
The most direct method of model creation was determined to be simply importing the bone scans into I-DEAS. The plan was to export the scans from DigiEdit in the dxf file format and then to use AutoCAD 14's igesout command-line function to convert the files into the igs file format. The files would finally be imported into I-DEAS using the import function (in IGES format). All of the bone manipulation (moving the bones into the spinal curvature and creating the intervertebral discs by boolean subtraction) would then be accomplished within the I-DEAS environment.
This plan, however, did not work. Most of the bone scans were too large for the I-DEAS software package to import. The conflict arose from a virtual memory conflict and from the Windows NT computer memory limitations. It should be noted that, while I-DEAS was able to import several of the undecimated bones, the memory needed by I-DEAS to hold an individual bone scan file in the I-DEAS workspace rendered the software incapable of manipulating the bone at all. It was also impossible to import more than one bone scan file into a single I-DEAS workspace. From these problems, Spinal Fusion determined that it was necessary to decimate the bone scans. An outline of this scheme can be seen in Fig. 4.1.
Figure 4.1 Direct Method Scheme
4.4.2 SCULPT Decimation
SCULPT, a software package used often in conjunction with the DigiEdit software in the Anthropology Department, has a built-in menu command for decimation in which the user inputs the approximate percentage of decimation. Spinal Fusion performed several decimations to determine the change in the appearance of the bone scans as a function of raising the percent of decimation. Through this process, it was determined that an approximate 75 percent decimation was the maximum decimation obtainable without significantly altering the appearance of the bone surface mesh. This inputted percentage of decimation produced differing actual percentages of decimation, due to the way SCULPT altered the mesh. The decimation process attempts to remove large changes in slope encountered in the juncture of the facets of the surface mesh. Most scans were decimated from 60-75 percent (actual decimation).
The first decimation scheme attempted involved using the SCULPT software package. The bone scans were exported from DigiEdit in the Wavefront object (obj) file format and opened in SCULPT. These scans were decimated and exported in the dxf file format. Again, the files were opened in AutoCAD 14. At this point, it was noted that many of the files now had incomplete surface meshes or corrupt surfaces. The decimated files that appeared to have complete meshes were exported (in IGES format) using the igesout command and then imported into I-DEAS. These files were determined by I-DEAS to be incomplete (free edges). Because decimation seemed to be the only way to allow I-DEAS to import the bone scans, another decimation program was located. An outline of this scheme of file manipulation can be seen in Fig. 4.2.
Figure 4.2 SCULPT Decimation Scheme
4.4.3 Visualization Toolkit (vtk) Decimate
The decimation scheme for the Visualization Toolkit Decimate (vtk Decimate) software was similar to the SCULPT decimation scheme, though an extra step was required. In order to use vtk Decimator the bone scans in dxf file format had to be converted to the stl (stereolithography) file format. This conversion was accomplished using a utility called Stl-Util. Once the files had been decimated in the vtk Decimate, they were converted back to the dxf file format. These files were then opened in AutoCAD 14 and exported using the igesout command-line function. As in the SCULPT decimation, it was determined that after the decimation/conversion process, the bone surface mesh was incomplete or corrupt. An outline of this decimation scheme can be seen in Fig. 4.3.
Figure 4.3 vtk Decimate/StlUtil Scheme
4.4.4 3-D Studio Max
After all of the previous attempts at file importation to I-DEAS had failed, Spinal Fusion tried one final scheme. This scheme was to import the dxf bone scans into 3-D Studio Max and complete all of the file manipulation within 3-D Studio Max. The bones were imported and positioned in the spinal curvature (see Fig. 4.1), though, due to time limitations (the end of the semester), the intervertebral discs were unable to be created. It is believed that these discs can be created by placing a simple cylinder between two vertebrae and performing a boolean subtraction of the adjacent vertebra. This theory, however, has not been tested. Assuming this process works, the final step for moving the spinal model into I-DEAS would be exporting it from 3-D Studio Max and importing it into I-DEAS. Fig. 4.4 and Fig. 4.5 depict the spine model in rendered 3-D Studio Max format and AutoCAD 14 dxf file format, respectively. Additionally, the file importation scheme for 3-D Studio Max can be seen in Fig. 4.6.
Figure 4.4 3-D Studio Max Rendered Spine
Figure 4.5 AutoCAD 14 dxf Spine
Figure 4.6 3-D Studio Max Importation Scheme
4.5 Literature Review
Spinal Fusion reviewed and compiled a large number of information sources throughout the semester. Some of these resources can be directly accessed on the team web page, some are included as references for this report, and the remaining sources can be found in the Spinal Fusion project notebook. This notebook can be found in W. R. Woolrich Laboratories (WRW) in room 316 at the University of Texas at Austin. In addition to research materials, the notebook contains copies of all of Spinal Fusion's semester work, including the midterm report, peer reviews, and group emails.
Resulting from the current inability to accurately model the spine, several alternative solutions are suggested by Spinal Fusion. One possible solution is the continuation of the 3-D Studio Max scheme of model importation. Another possibility is the exploration of commercial software packages capable of interfacing with I-DEAS.
5.1 3-D Studio Max Scheme
In order to develop the spine model in 3-D Studio Max, several steps are suggested:
After the complete spine is in the I-DEAS environment, material properties should be assigned to the different parts of the spine according to the data accumulated by A. Shirazi-Adl; see Table 2.1. FEM validation should be performed using the Persian Gulf demographic data, Table 2.2, and the helicopter crash ultimate loading data, Fig. 2.16. Once a verified model has been created, a single intervertebral disc should be replaced by a beam element to simulate spinal injury. The location of the injury should be varied along the spine, concentrating primarily on the lumbar and thoracic regions.
5.2 Commercial Solutions
Several commercial software packages were discovered while investigating finite-element models. These software packages include medical, hazardous event, and multipurpose models. The following sections detail available information for each of these systems. Spinal Fusion believes that it may be possible to circumvent all of the modeling difficulties encountered during this and all of the previous semesters by purchasing a commercially available software package that can accurately model the human spine and make modifications to this model to simulate injury. Also included in these descriptions are approximate current prices for the software.
5.2.1 Matrix 3-D
Matrix 3-D (formerly Medlink) is a solid-model generator, which interfaces directly with the I-DEAS software package. Bone scans taken directly from medical imaging hardware are processed using the Velocity medical imaging software (formerly Image Volume) and converted by Matrix 3-D into data sets that can be directly read by I-DEAS . In I-DEAS, loading and stress analysis could be accomplished after assembly of the components of the spine. The approximate price for this system obtained in a telephone interview with a customer service representative is $12500 for the software plus $1800 per year for maintenance. These combined software packages currently use I-DEAS Master Series 5.
As previously introduced, the ADAMS/Android is a complete human body model that is fully compatible with I-DEAS. The ADAMS/Android can be loaded and stressed directly in I-DEAS, though it is unclear whether individual components, such as the intervertebral discs, can be modified or excluded to simulate injury. The available literature claims a library of postures and the "ability to develop nonstandard body models or detailed body parts." Through a telephone interview with Chris Hetreed of Mechanical Dynamics, Inc., the approximate price was found to range between $2000-$6000 depending upon the types of components required by the user.
5.2.3 Articulated Total Body (ATB) Model
The Articulated Total Body model is an Airforce model that predicts human body dynamics in hazardous situations . As previously stated, this model runs under DOS or on a Silicon Graphics Workstation and is therefore independent of I-DEAS. This program may provide the required analysis on the spine, however, and should be considered a feasible (and inexpensive) approach to solving the modeling problem. The cost of the ATB is $1000 and can be purchased directly from the ATB Users' Group of the Airforce Systems Research Laboratories.
6.0 Organizational Overview
The organization of our project did not easily lend itself to a time schedule. This was because of the difficulties with the computational aspects of the project, specifically the undetermined status of the available IGES files. The components of our project are described in Fig. 6.1. The diagram shows our organization in tiers, beginning with the literature review progressing to the final finite-element model. This organizational approach will eventually produce the FEM, although, it will require additional semesters to complete.
Our group organization could be characterized as overlapping areas of responsibility. Jenna Bowling was group leader and was responsible for the administrative duties of the group. She was also a secondary researcher. Robin Kinsey was primary researcher, implying that nearly all research was her responsibility. Tony Chao was in charge of file retrieval/FEM modeling. These were the individualís primary duties, but each member was also obligated to supplement the other membersí contributions. Furthermore, each member was responsible for documentation of his/her tasks.
1. "Request For Proposal: Request for Cost Specific Proposal for Modeling Aviators with Diseased Spines for Safety in Aircraft Ejection," submitted by Lt. Col. Drew, Armstrong Laboratory, Brooks Air Force Base, Texas, July 1, 1996.
2. Coltman, J.W., Van Ingen, C., and Selker, F., "Crash-Resistant Crewseat Limit-Load Optimization Through Dynamic Testing with Cadavers," Simula, Inc., USAAVSCOM TR-85-D-11, Fort Eustis, VA, Jan. 1986.
3. "Anatomy of the Spine,"http://www.scoi.com/ spinant.htm.
4. Williams, Peter L., ed., Grayís Anatomy, 37th ed., Churchill Livingston, New York, 1989, pp. 268-493.
5. "Back Surface Anatomy, Vertebral Column and Musculature,"http://medic.med.uth.tmc.edu/Lecture /Main/back-surf.htm.
6. "Disc/Spine Assembly figure,"http://www.lowbackpain.com/spinal.html.
7. Goel, Vijay K., Park, Hosang, and Kong, Weizing, "Investigation of Vibration Characteristics of the Ligamentous Lumbar Spine Using the Finite Element Approach," Journal of Biomechanical Engineering, Vol. 116, Nov. 1994, p. 379.
8. "The Spine,"http://rothmaninstitute.com/spine/spinanat.htm.
9. Cantu, C. et al, "An Analysis of the Spine Subjected to Ejection Seat Loads," The University of Texas at Austin, 1997.
10. "The Ejection Site Facts Sheet,"http://www.bestweb.net/~kcoyne/ffacts.htm.
11. "ACES II,"http://22.214.171.124/NT/herker/ejection/aces.pic.html.
12. "The Ejection Site- The ACES II Seat: Tech Info.,"http://www.bestweb.net/~kcoyne/acesiitech.htm.
13. "Survival Features,"http://cust2.iamerica.net/blade/ survive.htm.
14. Osborne, Richard G. and Cook, Albert A., "Vertebral Fracture After Aircraft Ejection During Operation Desert Storm," Aviation, Space, and Environmental Medicine, Vol. 68, No. 4, April 1997, p. 338.
15. "Mathematical Modeling of the Cervical Spine,"http://www-ir.int.ethz.ch/research/elek/biomedizin/ niederer/pj.12.html.
16. Palumbo, Mark A. et al, "The Effect of Protective Football Equipment on Alignment of the Injured Cervical Spine," The American Journal of Sports Medicine, Vol. 24, No. 4, 1996, pp. 446-452.
17. "THOR Advanced Frontal Crash Test Dummy,"http://www-nrd.nhtsa.dot.gov/nnrd10/nrd12/thor/revintro.htm.
18. "Transom Jack,"http://www.transom.com/tj-humanfigure.html.
19. "Advanced Manikin Development,"http://www.al.wfafb.af.mil/cfb/manikin.htm.
21. "Articulated Total Body Model,"
22. Grant, S. and Lewis, K., "A Preliminary Report on a Study of Stress in Human Bone Structures," The University of Texas at Austin, 1995.
23. Frieling, J., et al, "Human Bone Stress Testing and Modeling," The University of Texas at Austin, 1996.
24. Hutchinson, W. and Littlefield, B., "Finite Element Model of the Human Spine," The University of Texas at Austin, 1996.
25. Haroz, C., et al, "A Model for Stress Analysis in Skeletal Structures," The University of Texas at Austin, 1996.
26. "Medlink,"http://www.sdrc.com/partners/solutions/ vendors/dynamic.
Request for Proposal from Brooks Air Force Base
Material Properties data referenced by Hutchinson/Littlefield
Shirazi-Adl, A., Ahmed, A. M., and Shrivastava, S.C., "A Finite Element Study of the Lumbar Motion Segment Subjected to Pure Sagittal Plane Moments," Journal of Biomechanics, Vol. 19, No. 4, 1986, p. 347.
Belytschko's Hypothesis (taken from Bone Crusher's Final Report)